Time Nick Message 00:52 magoo SEJeff: really like task warrior - just the thing I've needed 12:48 pdurbin check this upcoming desktop computer: HP t410 AiO (all in one). ARM Cortex-A8 processor ( http://infocenter.arm.com/help/index.jsp?topic=/com.arm.doc.ddi0344c/index.html ), uses only 13 watts of power, which is delivered power-over-ethernet: http://www.hp.com/united-states/campaigns/thin-client-solutions/t410.html#tab-graphics-specs . see also http://www.pcworld.idg.com.au/article/424196/new_hp_thin_client_cuts_power_cord/ 12:49 pdurbin "A standard desktop PC in a workplace environment consumes around 100 Watts not including the monitor, and the most efficient thin clients consume around 10 Watts. HP's t410 includes an 18.5in monitor for only three more Watts" 12:51 pdurbin "The other problem with the Intel instruction set is, as we saw, it's inefficient due to the 90/10 rule. So you have all these transistors sitting around not doing much because they just end up being the exception rather than the rule. Whereas this lean RISC design has all your transistors busy all the time, so that's efficient, too. You get a lot of use out of the power that the transistor is using." -- http://www.grc.com/sn/sn-252.htm 12:52 SEJeff But the tradeoff is performance 12:52 SEJeff arm is slow 12:53 SEJeff when compared to a modern intel processor such as a nehalum+ 12:53 pdurbin meh. i'm always at the command line :) 12:53 SEJeff exactly 12:53 SEJeff running perf tools :) 12:53 pdurbin should be fast enough. will it run fedora? as we've discussed fedora runs on ARM 12:54 SEJeff I'm not sure if all of fedora runs on arm 12:54 SEJeff It is still a non-primary arch for fedora 12:54 SEJeff So lots of ftbfs most likely 12:55 pdurbin Fails To Build From Source: http://en.wikipedia.org/wiki/FTBFS 12:56 pdurbin sure. but i'm still glad the fedora guys are working on ARM support 12:56 SEJeff indeed 13:18 pdurbin boooo, i'm running qemu-kvm-0.12.1 on centos 6, not 0.13.0 :( http://blog.wpkg.org/2010/10/04/resizing-qcow2-images/ i guess i'll have to convert this qcow2 image to raw to increase its size. . . 13:18 pdurbin would love to just run `qemu-img resize mydisk.qcow2 +5GB` per http://www.linux-kvm.com/content/how-resize-your-kvm-virtual-disk 13:24 SEJeff pdurbin, Question 13:24 SEJeff Have you ever backported newer stuff to RHEL/CentOS? If not, would you like a tutorial more or less? I do it all the time. 13:26 pdurbin well, sure. . . i've used the "clobber" repo from repoforge (or whatever it's called) before. in general, i try to stick with the generic upstream packages though. 13:28 SEJeff pdurbin, repoforge is bad 13:28 SEJeff The packages do a very good job of breaking your systems 13:28 pdurbin "rfx" they call it -- [suggest] Recent changes to RPMforge - http://lists.repoforge.org/pipermail/users/2010-November/018282.html via http://repoforge.org/faq/ . i voted for "clobber" 13:28 SEJeff It is very very easy to backport packages from Fedora to EL[56] using mock 13:29 pdurbin SEJeff: careful, one of the crimsonfu members is also a repoforge packager :) 13:29 SEJeff But if you'd prefer to just use raw disk images instead that works too. 13:29 SEJeff pdurbin, Well perhaps I should tell them that they break the GNOME yum auto-updates thing all the time :) 13:30 SEJeff for the gnome.org servers, which I help run 13:30 pdurbin SEJeff: yes, i backported the ikiwiki rpm from fedora for use in centos: http://yum.greptilian.com 13:30 pdurbin SEJeff: you should file an issue. i'm sure they'd appreciate it: https://github.com/repoforge/rpms/issues 13:30 SEJeff The only repo that is almost guaranteed to never break EL* is EPEL 13:30 SEJeff or if you're careful, stuff from fedora directly 13:31 SEJeff pdurbin, I actively steer people away from repoforge 13:31 SEJeff due to experiencing far too much breakage using it 13:31 pdurbin SEJeff: oh. you're one of those ;) 13:31 SEJeff *shrug* if the packages aren't high enough quality and break running systems 13:31 SEJeff It is a bad idea 13:32 pdurbin better not follow whorka's http://whorka.github.com/blog/2012/02/08/apache-mod_fastcgi-suexec-php-apc-rhel.html then :) 13:32 SEJeff one of those being defined as doesn't like to see broken deps ever 13:34 SEJeff pdurbin, But just saying. It is very easy to backport a onesy/twosey package and shouldn't be something you're afraid of if you want the functionality 13:34 pdurbin as we've discussed, there's much more work involved in getting packages in epel than repoforge: http://irclog.perlgeek.de/crimsonfu/2012-05-08#i_5559423 13:35 SEJeff pdurbin, Sure. There is more work because the demand higher quality packages. It shows :) 13:35 SEJeff You're just proving the point I was trying to make heh 13:36 pdurbin repoforge/rpmforge/dag has a much longer history than epel. in practice, i'm switching to epel 13:36 SEJeff However, once you do the initial work, it is very easy to get packages into epel. Even easier if they are already in Fedora. We've gotten perhaps a dozen or so packages in EPEL as hard and optional dependencies for salt. It wasn't very hard. 13:36 SEJeff dag was awesome back in the day 13:37 pdurbin i've been meaning to become a fedora/epel contributor 13:37 pdurbin SEJeff: would you consider qemu-kvm-0.12.1.2-2.209.el6_2.4.x86_64, qemu-img-0.12.1.2-2.209.el6_2.4.x86_64 and all related packages to be onesy/twosey? i'm not sure i would... 13:39 SEJeff I wouldn't build / install the kernel module for kvm 13:39 SEJeff but the userspace tools? Yes 13:39 SEJeff At worst, you use mockchain 13:39 SEJeff which is easy 13:40 pdurbin i need to get up to speed with mock 13:40 SEJeff pdurbin, If you ever have some time and want to let me know. I do it all the time as I'm the primary guy who manages our internal repos at work 13:40 pdurbin any handy links to howto's? 13:41 pdurbin if not, you could write up a crimsonfu blog post :) 13:41 SEJeff pdurbin, Oh and I wrote a really crappy wrapper for mock (and I mean really crappy) http://www.digitalprognosis.com/opensource/scripts/pkgbuild 13:41 SEJeff I wrote that for when we have to backport packages for GNOME 13:42 SEJeff Like when Redhat flat refused to backport the new sssd to RHEL 5.2 I think it was when we had issues which caused users to not have the ability to login to git.gnome.org (effectively stopped git contributions) 13:42 pdurbin SEJeff: ok, at least i see you're using mock in there. would probably be a good starting point for me 13:42 SEJeff I had to backport sssd for RHEL5 on the git server or switch back to pam_ldap, which is yucky in comparison 13:43 SEJeff pdurbin, the config is pretty straightforward. Look under /etc/mock. Edit the default config and then the distro specific one with your repos. 13:44 pdurbin huh, i just ran `qemu-img resize /var/lib/libvirt/images/pdurbin1-disk0.qcow +5GB` and it said "Image resized." . ls -l shows the same size though. . . 13:48 SEJeff holy netsplit batman! 13:49 pdurbin gparted should have a Requires: dejavu-lgc-sans-fonts 13:50 pdurbin yeah, i don't think that resize did anything. it lies with its "Image resized". no magic backport, i don't think. . . 13:52 pdurbin and a reboot didn't help either 13:55 SEJeff :/ 13:56 * pdurbin looks at http://pkgs.fedoraproject.org/gitweb/?p=qemu.git;a=blob;f=qemu.spec;hb=HEAD 13:56 pdurbin wow, lots of patches 13:56 SEJeff Also remember that most of the qemu upstream are redhatters 13:56 pdurbin sure 13:56 pdurbin is that what they call them? :) 13:57 SEJeff yes, /me has many friends there 13:58 SEJeff pdurbin, wow that is an insane number of patches though. You weren't kidding! 13:59 pdurbin still a onesy/twosey? :) 14:00 pdurbin at some point you just run your hypervisors on fedora rather than el6 if you want all the shiny 14:00 SEJeff pdurbin, 100 patches != 100 packages :) 14:01 SEJeff And I was just suggesting using the new qemu-image package. Does that require the new qemu-kvm kernel module? 14:01 SEJeff It might, but I'd be surprised 14:01 pdurbin no idea 14:01 SEJeff Avi K is good about keeping the kvm stuff backwards compat. I used to follow the kvm dev really closely 14:01 SEJeff Avi K == the guy who wrote kvm more or less 14:02 SEJeff http://avikivity.blogspot.com/ 14:06 magoo agoddard: you around? 14:09 pdurbin magoo: i don't see him on IM... 14:11 * pdurbin attempts a build of http://pkgs.fedoraproject.org/gitweb/?p=qemu.git;a=blob;f=qemu.spec;hb=35db8c0471c57821ed2b330304b69b4cedc551ea on el6 14:16 SEJeff pdurbin, Are you building from git or rebuilding a src rpm/ 14:16 pdurbin git 14:16 SEJeff hardcore 14:17 pdurbin do you have an srpm for this? i looked 14:17 SEJeff I normally grab the latest src rpm from koji or here: http://dl.fedoraproject.org/pub/fedora/linux/releases/ 14:17 SEJeff http://dl.fedoraproject.org/pub/fedora/linux/releases/test/17-Beta/Fedora/source/SRPMS/qemu-1.0-6.fc17.src.rpm 14:17 pdurbin so that's pretty new. . . too new? hmm 14:18 SEJeff I'm a bit confused 14:18 SEJeff That spec says version 0.13.0 14:18 SEJeff the srpm is 1.0.6 14:19 pdurbin i was thinking about going for a version that's just new enough to have the feature i need 14:19 SEJeff pdurbin, And if you want to grab the latest from rawhide: http://dl.fedoraproject.org/pub/fedora/linux/development/rawhide/source/SRPMS/q/qemu-1.0-17.fc18.src.rpm 14:19 SEJeff seems reasonable 14:20 pdurbin i wish there were a link from http://pkgs.fedoraproject.org/gitweb/?p=qemu.git back to https://admin.fedoraproject.org/pkgdb/acls/name/qemu . or maybe there is and i just don't see it 14:22 pdurbin i guess i could try http://kojipkgs.fedoraproject.org/packages/qemu/0.13.0/1.fc14/i686/qemu-img-0.13.0-1.fc14.i686.rpm via http://koji.fedoraproject.org/koji/buildinfo?buildID=201006 14:24 pdurbin whoops, http://kojipkgs.fedoraproject.org/packages/qemu/0.13.0/1.fc14/x86_64/qemu-img-0.13.0-1.fc14.x86_64.rpm i mean (not i686) 14:34 pdurbin bah! ran `rpm -Uvh qemu-img-0.13.0-1.fc14.x86_64.rpm --nodeps` and tried a resize on qcow2 but it still didn't seem to work 14:35 pdurbin actually, hmm! qemu-info is showing the virtual size increasing... 14:36 pdurbin er, `qemu-img info` i mean 14:38 SEJeff pdurbin, Sparse image file? 14:39 pdurbin i'm probably just being dumb. probably need to boot into Parted Magic like I've done before: http://partedmagic.com 14:51 pdurbin yup. just me being dumb. plenty of unallocated when i run gparted from within parted magic. i should have followed my own doc from the start :) 15:36 pdurbin good old gparted 15:44 pdurbin i have all kinds of corrections to the noise i was making above 15:50 pdurbin wow, somehow i didn't realize `virsh setmem myvm 2097152` would have an immediate affect on a running vm. MemTotal in /proc/meminfo is suddenly bigger 15:50 pdurbin ventz: well, truthfully, you told me yesterday this would work but i only half believed you :) 16:12 pdurbin ok, so corrections to the qemu resize business above 16:13 pdurbin stock el6 qemu-img resizes qcow2 images just fine 16:13 pdurbin never never believe the size reported by `ls -l`. (i should have known better). always run `qemu-img info` instead 16:14 pdurbin to actually do the resize, reboot with a Parted Magic iso and use gparted 16:14 pdurbin or some fancier command line only way if you have the fu, i suppose 16:14 SEJeff parted isn't that hard to use from the cli 16:14 pdurbin SEJeff: yeah? have a link to a nice howto? 16:14 SEJeff I learned from the manpage :) 16:14 SEJeff Let me look one up 16:15 SEJeff http://www.cyberciti.biz/tips/fdisk-unable-to-create-partition-greater-2tb.html http://www.gnu.org/software/parted/manual/html_chapter/parted_2.html 16:15 SEJeff That manual is actually really good 16:16 pdurbin cool. thanks 16:16 pdurbin but you have to boot off some iso or into rescue mode, right? 16:16 pdurbin you can't just run parted on a running system, can you? 16:19 pdurbin that is to say, you have to unmount the file system you're working on... 16:20 pdurbin that's what i've been doing anyway. booting off the parted magic iso 16:22 SEJeff pdurbin, Oh I thought you were doing this for vms 16:22 pdurbin i am. these are vms 16:22 SEJeff If you're using lvm you can grow things live 16:22 pdurbin kvm 16:22 SEJeff lvm + ext4 I believe 16:22 ventz pdurbin: using resize2fs is very easy 16:22 SEJeff ++ 16:22 pdurbin well, the vm i was working on wasn't using lvm 16:23 SEJeff You need the filesystem and underlying partitions to grow 16:23 SEJeff If the filesystem doesn't fill the entire partition, then you just use resize2fs like ventz mentioned without worrying. Otherwise, yes the way you did it works 16:23 ventz my favorite for lvm: 16:23 ventz lvdisplay 16:23 ventz swapoff /dev/VolGroup00/LogVol01 16:23 ventz lvresize -L -4GB /dev/VolGroup00/LogVol01 16:23 ventz lvresize -L +4GB /dev/VolGroup00/LogVol00 16:23 ventz resize2fs -p /dev/VolGroup00/LogVol00 16:24 ventz mkswap /dev/VolGroup00/LogVol01 16:24 ventz swapon /dev/VolGroup00/LogVol01 16:24 ventz # fixing centos one day at a time ;) 16:26 pdurbin ok, but for non-lvm, i think i need to reboot at least, for the ubuntu vm to know that i made the qcow2 image bigger 16:26 SEJeff pdurbin, There might be a trick you can do 16:26 pdurbin SEJeff: yeah, probably there is 16:26 SEJeff There is a way to force the kernel to re-read the partition table. Hold on 16:26 pdurbin ventz: thanks for the lvm fu 16:27 SEJeff pdurbin, partprobe 16:27 SEJeff partprobe /dev/XXX 16:27 pdurbin right. that makes sense 16:27 pdurbin will have to try that in my test vm 16:28 SEJeff pdurbin, I'm pretty sure partprobe calls or does the equivalent of: blockdev --rereadpt /dev/sda 16:28 pdurbin ok 16:28 SEJeff blockdev was what I was originally thinking of 16:29 pdurbin i definitely seem to have had to reboot the vm to increase the max ram 16:29 SEJeff That most likely depends on the kernel 16:29 SEJeff and kvm version 16:30 pdurbin "error: Unable to change MaxMemorySize , error: Requested operation is not valid: cannot resize the maximum memory on an active domain" from `virsh setmaxmem myvm 2097152` 16:30 pdurbin qemu-kvm-0.12.1.2-2.209.el6_2.4.x86_64 kernel-2.6.32-220.7.1.el6.centos.plus.x86_64 17:34 SEJeff Might be relevant in this channel: http://jpmens.net/2010/10/29/alternative-dns-servers-the-book-as-pdf/ 17:34 SEJeff pdurbin, It has a bit about powerdns too ^^ 17:35 pdurbin huh. very cool 17:59 pdurbin "You just need to do apt-get install <packagename>. This will upgrade only that single package." -- http://askubuntu.com/questions/44122/upgrade-a-single-package-with-apt-get/44124#44124 ubuntu is so odd... 19:11 pdurbin trying to remember NX :) 19:46 pdurbin looking at http://perldoc.perl.org/perlfaq8.html#How-do-I-keep-my-own-module/library-directory? via http://stackoverflow.com/questions/1366902/why-does-installing-certain-cpan-modules-require-root-privilege/1370065#1370065 19:47 pdurbin cpan> o conf makepl_arg INSTALL_BASE=/mydir/perl 19:47 pdurbin i guess that's what i'm looking for 20:04 pdurbin that config option seems to work: http://pastie.org/3932312